Pile settlement prediction applying hybrid ALO-SVR and BBO-SVR approaches

Pile settlement (SP) socketed to rock has taken substantial regard. Despite introducing some design techniques of SP, employing the novel and workable prediction model with acceptable prediction performance is pivotal. The basic target of the present paper is to find out the applicability of applying two hybrid biogeography-based support vector regressions (BBO-SVR) and ant lion optimization-support vector regression (ALO-SVR) in predicting the SP in the Klang Valley Mass Rapid Transit (KVMRT) project constructed and operated in Kuala Lumpur, Malaysia. To apply the hybrid methods, the records of pile driving analyzer tests, piles, and earth’s properties were considered. Comparing the recorded SP with predicted values by ALO-SVR and BBO-SVR are supplied that the developed models have R2 larger than 0.9994. Considering all statistical evaluators, it could be concluded that both hybrid models are really capable of predicting SP. However, ALO-SVR represents a higher ability in determining the optimal value of the SVR parameters than BBO-SVR.